Titulo Estágio
Intent-Based Interaction for Anomalous Network Traffic Detection in Cloud-Native Environments
Áreas de especialidade
Sistemas Inteligentes
Engenharia de Software
Local do Estágio
Rua Dom João Castro n.12, 3030-384 Coimbra, Portugal
Enquadramento
In cloud-native environments, where applications are composed of dynamic microservices and complex network interactions, ensuring network security and visibility is increasingly difficult. Modern solutions often include frameworks that automatically detect and classify anomalous network traffic to help identify threats, misconfigurations, or performance issues. However, these systems typically require manual interaction with APIs, dashboards, or logs, which limits their usability for non-expert users or during high-pressure incident response.
To bridge this gap, intent-based interfaces enable users to interact with systems using natural language, abstracting the complexity of underlying queries. This paradigm allows operators to state what they want (e.g., "Show me suspicious traffic from the frontend service in the last hour") and receive results aligned with their intent, presented in both text and visual dashboards.
DeepGuardian, designed by OneSource, is a framework that leverages Machine Learning (ML) models to detect and classify anomalies in network traffic. This framework allows real-time detection and classification of outbound and inbound network traffic from cloud-native applications.
Objetivo
The goal of this internship is to develop an intent-based interface over DeepGuardian. The solution will include:
1. Analyze existing intent recognition techniques across various fields such as Natural Language Processing (NLP) and Human-Computer Interaction (HCI);
2. Develop intent-based models that recognize and account for user or system intent, allowing for more context-aware and personalized decision-making processes;
3. Evaluate the effectiveness of these models in real-world applications, comparing them to traditional models in terms of accuracy, personalization, user trust, and security;
4. Implement a prototype system that incorporates intent-based modeling, testing its impact on DeepGuardian framework;
Plano de Trabalhos - Semestre 1
1. Project setup and familiarization: Understand the DeepGuardian framework, its architecture, APIs and output formats;
2. Literature and technology review: Explore intent recognition techniques in NLP and HCI, and research chatbot frameworks;
3. Intent Schema and Use Case Design: Identify and define key user intents, define required entities and design typical interaction flows between user, chatbot and DeepGuardian;
4. System Architecture Design: Design the full system: chatbot layer and intent processor;
5. Intermediate Report: Write the first draft of the thesis, summarizing the research problem, objectives, related work, and preliminary findings.
Plano de Trabalhos - Semestre 2
1. Model Development: Develop intent-based models that recognize and adapt to user or system intent, ensuring that these models can enhance personalization, security, and interpretability.
2. Prototype Development: Build a minimal viable product (MVP) chatbot handling basic intents and integrate with DeepGuardian;
3. System Testing and refinement: Perform end-to-end testing with variety of intents and queries, improve NLP model accuracy, error handling and intent detection based on test results;
4. Comparative Evaluation: Compare intent-based interaction with traditional use of DeepGuardian; Measure usability, speed of insight, trust and interpretability;
5. Final Thesis Writing: Compile the research findings, methodology, and evaluation results into the final version of the master’s thesis.
Condições
The trainee will have all the necessary conditions to carry out the planned tasks, being integrated into the research and development teams within European research projects in which OneSource is involved.
Orientador
Jorge Diogo Gomes Proença
jorge.proenca@onesource.pt 📩